The present invention relates generally to transistor devices and, more particularly, to pass transistor structures and methods.
An important capability in modern electronic systems is the capability for making programmable interconnections between logic components or logic blocks. This capability is provided by components such as Field Programmable Gate Arrays (FPGAS) and other programmable interconnect logic and circuits. Consequently, it is important to find methods to improve performance of these devices, reduce their power consumption and simplify their structures and design.
FIG. 1 is a schematic representation of a typical prior art structure 100 that included: prior art logic block 101; prior art logic block 103; and prior art transmission gate 105, coupled between prior art logic block 101 and prior art logic block 103. As shown in FIG. 1, prior art logic block 101 included input node 111 and output node 113. Like prior art logic block 101, prior art logic block 103 also included an input node 119 and output node 121. Prior art logic blocks 101 and 103 typically comprised any one of numerous devices well known to those of skill in the art such as single transistors, inverters, latches, any one of several gates, or any other logic or memory devices. Both prior art logic blocks 101 and 103 were provided with prior art first supply voltage 131 and prior art second supply voltage 133. Since prior art logic blocks 101 and 103 were typically standard CMOS, the prior art supply voltage was typically on the order of two volts or more. Consequently, prior art first supply voltage 131 was typically 2.0 volts and prior art second supply voltage 133 was typically ground.
As seen in FIG. 1, prior art structure 100 also included prior art transmission gate 105 for selectively, and programmably, connecting prior art logic blocks 101 and 103. Prior art transmission gate 105 included first node 115, coupled to output node 113 of prior art logic gate 101, and second node 117, coupled to input node 119 of prior art logic block 103. Prior art transmission gate 105 also included: prior art NFET 107 with gate 107G and prior art PFET 109 with gate 109G. Prior art NFET 107 and prior art PFET 109 were typically coupled together as shown in FIG. 1, to allow transmission of both digital one and digital zero values as discussed in more detail below.
Prior art transmission gate 105 had three significant drawbacks. First, the addition of prior art transmission gate 105 added significant resistance to the path between output node 113 of prior art logic block 101 and input node 119 of prior art logic block 103, i.e., prior art transistors 107 and 109 each added a resistance in series to the path between prior art logic block 101 and prior art logic block 103. Second, prior art transmission gate 105 was a relatively complex structure requiring the use of at least two transistors, 107 and 109. Third, prior art transmission gate 105 added significant parasitic capacitance to the path between output node 113 of prior art logic block 101 and input node 119 of prior art logic block 103. The added resistance and parasitic capacitance meant decreased performance of prior art structure 100 and increased power dissipation. The performance reduction due to the addition of prior art transmission gate 105, and prior art transistors 107 and 109, could be partially offset in the prior art by increasing the size of prior art transistors 107 and 109 relative to the size of the transistors making up logic blocks 101 and 103 (not shown). However, even a ten fold increase in relative size of prior art transistors 107 and 109, compared to the transistors in logic blocks 101 and 103, would still typically yield a decrease in performance of more than ten percent. This was still a very significant performance loss.
A theoretical way to minimize the decrease in performance of prior art structure 100 due to the addition of prior art transistors 107 and 109 would be to drive prior art transistors 107 and 109 at a higher voltage than the voltage driving logic blocks 101 and 103. However, in practice, to actually make any significant difference in the performance, i.e., to significantly decrease the resistance added by prior art transistors 107 and 109, the supply voltages of prior art gating transistors 107 and 109 would need to be multiples, and preferably an order of magnitude, larger than the differential between first supply voltage 131 and second supply voltage 133. However, as noted above, in standard CMOS, the voltage differential between first supply voltage 131 and second supply voltage 133 is on the order of two volts. Standard transistors typically cannot tolerate more than about 2.5 volts in 0.25 micron technology with 50 angstroms of gate oxide, consequently, the voltage differential required to significantly decrease the added resistance of prior art transistors 107 and 109 could not be withstood by standard CMOS transistors, over time, and prior art transistors 107 and 109 would eventually break down.
The relative complexity of prior art transmission gate 105, i.e., the need for two prior art transistors 107 and 109, arose from the relatively high threshold voltages of prior art transistors 107 and 109 and from the well known body effect. As a result, NFETs, such as prior art transistor 107, could pass a digital zero relatively well but could not pass a digital one efficiently. On the other hand, PFETs, such as prior art transistor 109 could pass a digital one relatively well but could not pass a digital zero well. Consequently, in prior art transmission gates, such as prior art transmission gate 105, an NFET, such as prior art transistor 107, was included to pass digital zeros while a PFET, such as prior art transistor 109 was included in transmission gate 105 to pass digital ones. This coupling of NFETs and PFETs performed reasonably well in prior art transmission gates 105. However, the use of two transistors 107 and 109 meant increased parasitic capacitance, increased area, and increased circuit complexity, with more elements to potentially fail.
What is needed is a method and apparatus for providing a transmission gate function between two logic blocks that is more efficient, in terms of lowering added resistance, lowering added parasitic capacitance and in terms of circuit complexity and size.
According to one embodiment of the invention, low voltage programmable logic structures are provided that include low voltage logic blocks comprised of low threshold transistors. The low voltage logic blocks are provided with low supply voltages. According to one embodiment of the invention, the low voltage logic blocks are separated by pass transistors. Since the logic blocks of the invention are low voltage, the pass transistors of the invention can be overdriven on, i.e., provided gate to source voltages (Vgs) significantly larger than the programmable logic structure""s supply voltage, without causing the destruction of the pass transistor. For instance, in one embodiment of the invention, the programmable logic structure""s supply voltage is 200 millivolts while the gate to source voltage (Vgs) of the pass transistor, when the transistor is overdriven on, is on the order of 2.0 volts. This is in direct contrast to prior art transmission gates that had to be driven at essentially the same Vgs as the supply voltage to avoid transistor breakdown. By overdriving the pass transistors of the invention, the resistance added to the structure by the pass transistors, i.e., the resistance added by coupling the pass transistor between logic blocks is decreased significantly without resorting to increasing the size of the pass transistors.
In addition, if even less resistance is desired, the size of the pass transistors of the invention can be increased and the pass transistors can still be overdriven according to the invention. Consequently, using the invention, there is an approach for resistance reduction available that was not available in the prior art and can provide orders of magnitude decrease in the added resistance along with a decrease in added parasitic capacitance. Therefore, the invention can provide the advantages of a programmable interconnection without the large performance penalty associated with the prior art.
In addition, if even less resistance is desired, according to the invention, the pass transistors can be low threshold transistors. In this embodiment the low threshold pass transistors are overdriven on to reduce resistance as described in detail below and overdriven off to reduce leakage current and further isolate the logic blocks.
In addition, as discussed in more detail below, since the pass transistors of the invention are overdriven on, a single NFET can pass a digital zero or a digital one efficiently, or a single PFET can pass a digital one or digital zero efficiently. This situation is further improved when the pass transistors are low threshold pass transistors. Consequently, using the method and structure of the invention, a single pass transistor can be used as opposed to the two transistor structure of the transmission gates of the prior art. Consequently, a programmable logic structure designed according to the principles of the method and the structure of this embodiment of the invention has better performance, reduced parasitic capacitance, is simpler and less expensive to implement than prior art structures.
It is to be understood that both the foregoing general description and following detailed description are intended only to exemplify and explain the invention as claimed.